Actions and Objects Pathways for Domain Adaptation in Video Question Answering

Fuente: arXiv
Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Mohamud, Safaa Abdullahi Moallim, Jung, Ho-Young
Format: Preprint
Veröffentlicht: 2024
Schlagworte:
Online-Zugang:
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
_version_ 1866912137201319936
author Mohamud, Safaa Abdullahi Moallim
Jung, Ho-Young
author_facet Mohamud, Safaa Abdullahi Moallim
Jung, Ho-Young
contents In this paper, we introduce the Actions and Objects Pathways (AOPath) for out-of-domain generalization in video question answering tasks. AOPath leverages features from a large pretrained model to enhance generalizability without the need for explicit training on the unseen domains. Inspired by human brain, AOPath dissociates the pretrained features into action and object features, and subsequently processes them through separate reasoning pathways. It utilizes a novel module which converts out-of-domain features into domain-agnostic features without introducing any trainable weights. We validate the proposed approach on the TVQA dataset, which is partitioned into multiple subsets based on genre to facilitate the assessment of generalizability. The proposed approach demonstrates 5% and 4% superior performance over conventional classifiers on out-of-domain and in-domain datasets, respectively. It also outperforms prior methods that involve training millions of parameters, whereas the proposed approach trains very few parameters.
format Preprint
id arxiv_https___arxiv_org_abs_2411_19434
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Actions and Objects Pathways for Domain Adaptation in Video Question Answering
Mohamud, Safaa Abdullahi Moallim
Jung, Ho-Young
Computer Vision and Pattern Recognition
Computation and Language
In this paper, we introduce the Actions and Objects Pathways (AOPath) for out-of-domain generalization in video question answering tasks. AOPath leverages features from a large pretrained model to enhance generalizability without the need for explicit training on the unseen domains. Inspired by human brain, AOPath dissociates the pretrained features into action and object features, and subsequently processes them through separate reasoning pathways. It utilizes a novel module which converts out-of-domain features into domain-agnostic features without introducing any trainable weights. We validate the proposed approach on the TVQA dataset, which is partitioned into multiple subsets based on genre to facilitate the assessment of generalizability. The proposed approach demonstrates 5% and 4% superior performance over conventional classifiers on out-of-domain and in-domain datasets, respectively. It also outperforms prior methods that involve training millions of parameters, whereas the proposed approach trains very few parameters.
title Actions and Objects Pathways for Domain Adaptation in Video Question Answering
topic Computer Vision and Pattern Recognition
Computation and Language
url https://arxiv.org/abs/2411.19434